PM2005: Discrepancies in ASQ Measurements of Infant Social and Behavioral Health

Background: Early social emotional development provides the foundation for later physical and mental health, and the AAP recommends screening for early signs of risk (1). Screens for social emotional concerns include specialized screens like the Brief Early Childhood Screening Assessment (B-ECSA) (2) or global developmental screens, like the Ages and Stages Questionnaire-3 (ASQ) (3). The B-ECSA describes symptoms of social emotional dysregulation and has been rigorously validated using the highest standard of validation, a diagnostic interview. The ASQ describes various developmental milestones across multiple domains at specific time points. The CDC recently revised developmental surveillance checklists, with social emotional development having limited normative data, raising questions about this milestone-based screening (4).

Using a large dataset, we examine the comparative validity of the B-ECSA and ASQ-3 in identifying children with social emotional concerns and predictive validity of the ASQ, particularly the personal social scale, in identification of children in need of additional evaluation.

Methods: Clinical, screening and demographic information for 1456 18-month-old whose parents completed a B-ECSA was extracted from a national screening platform (CHADIS). When available, ASQ results from 2 to 24 months of age as well as adverse social determinants of health factors were extracted from CHADIS. B-ECSA and ASQ results were compared using Chi-Square. When multiple time points existed for ASQ, predictive validity was examined using Chi-Square. Data analysis was completed in SPSS-26.

Results: At 18 months, although ECSA and ASQ personal-social scale are associated (x2(1)= 4.9; p<.03), a positive ASQ personal social scale only identified 5.8% of children who screened positive on the ECSA. Specificity was 97.9%. Children with a positive ASQ personal social screen had a mean ECSA score of 4.5, substantially lower than the clinical cut-off of 9. ASQ personal-social predictive validity between each consecutive well child visit (6, 9, 12, 15, and 18 months) was low, with fewer than 36.4% of children who screened positive continuing to screen positive at the next well child visit.

Conclusion: Overall, this study identifies limited concurrent association between the personal social subscale of a widely used measure and a validated measure specific for emotional well-being as well as limited predictive validity of the ASQ-3 personal social scale over time. The difference between the ASQ personal social scale and the ECSA raise important clinical questions about equivalence and domains measured. The limited predictive validity of the ASQ personal social scale over time raises additional questions about the milestone selection and how results should be interpreted in pediatric primary care. Further research into the long-term developmental trajectory of patients with positive and negative screens is needed to ensure early childhood social emotional measures are accurate and actionable.